***Readme for 'Expectation Formation Following Large Unexpected Shocks' by Baker, McElroy and Sheng

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
***Empirical Results
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

***Natural Disaster data
1) disaster_data.dta - This data contains information about the various natural disaters used
in the paper as unexpected shocks.

***Newspaper Code
2) Constuct Newspaper Data and Figure 1.do - This code builds the newspaper data in the days
surrounding each natural disaster based on data scraped from Access World News. Newspaper article
counts are included in file disaster_news_g7.csv and disaster_news_worldwide.csv. The code also
generates Figure 1, showing the average jump in news coverage surrounding a disaster.

***Main Forecast Code
3) Aggregate Forecasts Tables.do - This code compiles and cleans the aggregated mean/std dev of 
forecasts from Consensus Economics for the full set of countries and merges in data regarding the 
disasters and newspaper articles. The code also generates Tables 1 (summary stats), 2, 3, 2b, and 
3b (2b and 3b are in the Appendix).

4) Individual Forecasts Figures+Tables.do - This code compiles and cleans the individual forecast
data from Consensus Economics for the available G7 countries. It also merges in data regarding
the disasters and newspaper articles. The code generates Tables 4-6 as well as Figures 2-3.

Forecast data are proprietary, and can be purchased from Consensus Economics.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
***Simulation + Simulation Results
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

1) ScriptReStat.r: R script to generate simulations and all graphs
2) FunctionsReStat.r: all R functions needed  
3) IRrevision.RData: R workspace with simulations used in paper